Problem with InfinitePolynomialRing

[Martin R]

The following is giving me a headache.  In a fresh session, do

sage: R.<a> = InfinitePolynomialRing(SR)
sage: p = a[0].polynomial()
sage: R(p)

...

RuntimeError: Symbolic Ring still using old coercion framework

I tried all day to make sense of the element_constructor of the InfinitePolynomialRIng (which uses sage_eval to interpret the repr), but I failed.

Martin

[julian...@fsfe.org]

Hi Martin,

On Tuesday, April 2, 2024 at 3:17:19 PM UTC+3 Martin R wrote:

I tried all day to make sense of the element_constructor of the InfinitePolynomialRIng (which uses sage_eval to interpret the repr), but I failed.

That string eval is quite curious. I have no clue why it's necessary. However, isn't the underlying issue that the symbolic ring has no gens (and no ngens, and no gens_dict)?  I have no clue what the correct implementation could look like but all these just throwing a RuntimeError looks quite wrong to me. Any other implementation seems to improve the situation:

sage: class SR_(sage.symbolic.ring.SymbolicRing):
....:     def gens_dict(self): return {}
....:
sage: SR = SR_()

sage: R.<a> = InfinitePolynomialRing(SR)
sage: p = a[0].polynomial()
sage: R(p)

a_0

Alternatively,

sage: class SR_(sage.symbolic.ring.SymbolicRing):
....:     def gens_dict(self): raise ValueError() # caught by the infinite polynomial ring

[...]

sage: R(p)

ValueError: cannot convert a_0 into an element of Infinite polynomial ring in a over Symbolic Ring - no conversion into underlying polynomial ring

I'm also no expert when it comes to the symbolic ring. But I'm on Zulip if you want to investigate together what's going on.

julian